Reliability of last menstrual period recall, an early ultrasound and a Smartphone App in predicting date of delivery and classification of preterm and post-term births.

Background: A reliable expected date of delivery (EDD) is important for pregnant women in planning for a safe delivery and critical for management of obstetric emergencies. We compared the accuracy of LMP recall, an early ultrasound (EUS) and a Smartphone App in predicting the EDD in South African pregnant women. We further evaluated the rates of preterm and post-term births based on using the different measures.
Methods: This is a retrospective sub-study of pregnant women enrolled in a randomized controlled trial between October 2017-December 2019. EDD and gestational age (GA) at delivery were calculated from EUS, LMP and Smartphone App. Data were analysed using SPSS version 25. A Bland-Altman plot was constructed to determine the limits of agreement between LMP and EUS.
Results: Three hundred twenty-five pregnant women who delivered at term (≥ 37 weeks by EUS) and without pregnancy complications were included in this analysis. Women had an EUS at a mean GA of 16 weeks and 3 days). The mean difference between LMP dating and EUS is 0.8 days with the limits of agreement 31.4-30.3 days (Concordance Correlation Co-efficient 0.835; 95%CI 0.802, 0.867). The mean(SD) of the marginal time distribution of the two methods differ significantly (p = 0.00187). EDDs were < 14 days of the actual date of delivery (ADD) for 287 (88.3%;95%CI 84.4-91.4), 279 (85.9%;95%CI 81.6-89.2) and 215 (66.2%;95%CI 60.9-71.1) women for EUS, Smartphone App and LMP respectively but overall agreement between EUS and LMP was only 46.5% using a five category scale for EDD-ADD with a kappa of .22. EUS 14-24 weeks and EUS < 14 weeks predicted EDDs < 14 days of ADD in 88.1% and 79.3% of women respectively. The proportion of births classified as preterm (< 37 weeks) was 9.9% (95%CI 7.1-13.6) by LMP and 0.3% (95%CI 0.1-1.7) by Smartphone App. The proportion of post-term (> 42 weeks gestation) births was 11.4% (95%CI 8.4-15.3), 1.9% (95%CI 0.9-3.9) and 3.4% (95%CI 1.9-5.9) by LMP, EUS and Smartphone respectively.
Conclusions: EUS and Smartphone App were the most accurate to estimate the EDD in pregnant women. LMP-based dating resulted in misclassification of a significantly greater number of preterm and post-term deliveries compared to EUS and the Smartphone App.